Cutting Edge: Rapid and Efficient In Vivo Cytotoxicity by Cytotoxic T Cells Is Independent of Granzymes A and B

Mice, virus, and cell lines

Female C57BL/6 and gene knockout (KO) mice were obtained from the specific pathogen-free facility at the John Curtin School of Medical Research (JCSMR) in Canberra, Australia or the Peter MacCallum Cancer Institute in East Melbourne, Australia (cathepsin C-deficient (B6.CatC^−/−) mice) and used according to institutional animal experimentation approval. The influenza virus A/PR8 and Hampstead egg (HE)-ECTV were grown and assayed as described previously (15, 18).

In vivo killing assays

For in vivo Tc cell assays, CD45.2 mice were infected i.v. with 10⁵ PFU of HE-ECTV or 10³ hemagglutination units of influenza A/PR8 virus. Six to 7 days later, immunized mice were injected i.v. with a 1:4 mixture (2 × 10⁷ total cells) of control-peptide-pulsed, or ECTV Tc cell determinant-pulsed (EV-K^b, H-2K^b-restricted TSYK FESSV) (19), and cell tracker-labeled (7-hydroxy-9H-(1,3-dichloro-9,9-dimethylacridin-2-one) or DDAO; 10 μM) CD45.1 congenic splenocytes. Alternatively, control peptide-pulsed CFSE^low and cognate peptide-pulsed (influenza virus H-2D^b-restricted ASNENMETM; NP-D^b) CFSE^high splenocytes from Fas-negative mice were injected into primed syngenic gene KO mice. At different time points, spleens, popliteal lymph nodes, and livers of recipient mice were harvested and donor target cells (CD45.1⁺ or CFSE⁺) were enumerated by FACS.

FACS

Splenocytes, lymph node, and liver cells were incubated in 7-amino-actinomycin D and stained for CD8α (BD Pharmingen) and tetramers comprising ECTV K^b-tetramer (produced in the Australian Cancer Research Foundation (ACRF) Biomolecular Resource Facility, JCSMR, Canberra, Australia). Cells were then fixed in 2% fresh paraformaldehyde, permeabilized with 0.5% saponin, and stained with polyclonal rabbit anti-gzmA or anti-gzmB antiserum, followed by Alexa Fluor 647-conjugated rat anti-rabbit IgG (Molecular Probes). Cells were then analyzed on a FACSCalibur flow cytometer (BD Biosciences) and on WEASEL FACS software (Walter and Eliza Hall Institute of Medical Research, Melbourne, Australia). The percentage (%) of specific target cell elimination was calculated as percent specific loss of cognate target cells = 100 − 100/(number of recovered control cells) × (number of recovered specific target cells), normalized to the ratio of the injected cell mix.

Semiquantitative real-time PCR for perf and gzm gene expression

Total mRNA was isolated from 5 × 10⁵ CD8⁺ cells, positively enriched by autoMACS according to the manufacturer’s protocol (Miltenyi Biotec), reverse transcribed, and assayed with TaqMan gene expression assays (Applied Biosystems) on a 7900HT fast real-time PCR system (Applied Biosystems, at the ACRF Biomolecular Resource Facility, JCSMR). The copy number obtained for each gene of interest was normalized to the copy number of GAPDH mRNA (= 1).

Mousepox-immune Tc cells eliminate cognate target cells independently of gzmA and gzmB

To determine the short-term in vivo cytotoxic potential of gzmA×B^−/− Tc cells, we adapted a frequently used in vivo killing assay in which cell tracker-labeled target cells, pulsed either with control peptide or a cognate Tc cell determinant, are coinjected i.v. into immunized recipients and recovery in several organs is tested at several time points after injection (20). We infected wild-type (WT) or gzmA×B^−/− mice with the attenuated ECTV strain HE-ECTV and used the immunodominant ECTV Tc cell determinant as a cognate peptide. Similar to the kinetics of in vivo target killing reported by others for mice infected with lymphocytic choriomeningitis virus (LCMV) (21, 22) and other viruses (23), in vivo elimination by ECTV-primed WT mice is extremely rapid, being measurable in both spleen and lymph nodes only minutes after transfer (Fig. 1⇓). Elegant mathematical algorithms have been proposed that model the observed rapid target cell killing in LCMV assays and suggest a Tc cell killing rate of 2–4/min (24) or target cell half-lives of 2–14 min (25). Because, in vitro Tc cells from ECTV-immune gzmA×B^−/− mice are as kinetically deficient at inducing apoptotic features (phosphatidylserine exposure and caspase 3 activation; J. Pardo, M. Regner, A. Mullbacher, manuscript in preparation) in a range of target cells as those from LCMV-immune mice (8), we expected that the elimination of cognate target cells in vivo would be delayed or abrogated. Surprisingly, gzmA×B^−/− mice cleared cognate targets as efficiently and rapidly as WT mice (Fig. 1⇓). This suggests that neither of the two dominantly expressed Tc cell granzymes is required for efficient in vivo clearance of cognate target cells.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Rapid and efficient in vivo target cell killing by antiviral Tc cells. WT or gzmA×B^−/− mice (CD45.2) were infected with ECTV and 6 days later injected with splenocytes from naive congenic (CD45.1) mice, either labeled with cell tracker DDAO and the immunodominant Tc cell determinant from ECTV (EV-K^b) or a control (Ctrl) peptide (NP-D^b; cell ratio 4:1). A, After the indicated time points, spleens and popliteal lymph nodes were removed and the ratios of adoptively transferred target cells in these organs were determined. B, Strategy for gating donor target cells. C, Kinetics of specific elimination of cognate target cells in spleen (top graph) and lymph nodes (bottom graph) of virus-immune WT or gzmA×B^−/− mice, respectively. Naive WT and gzmA×B^−/− recipients are shown for 150 min. Data shown are means with SD from three mice per time point and are representative of two separate experiments.

In the study originally describing this in vivo killing assay, it was noted that the elimination of target splenocytes was perf-independent, suggesting that it may not be mediated by the granule exocytosis pathway (20). However, since then others have shown strong perforin-dependency in the same LCMV system (21) (albeit in memory rather than primary responses) as well as in other infection models (23). Although we cannot account for the discrepancies in the LCMV system, we found that the clearance of cognate cells in ECTV infection was strongly perf dependent but Fas independent, because WT and gzm-deficient but not perf-deficient (Fig. 2⇓A) mice could eliminate Fas-deficient targets with similar efficiency (Fig. 2⇓A) and kinetics (not shown) as those of WT mice. Importantly, all gene KO mice tested had generated similar magnitudes of the peptide-specific Tc cell responses investigated, because all harbored similar proportions (Fig. 2⇓B) and total numbers (not shown) of tetramer-binding CD8⁺ splenic and popliteal lymph node (not shown) T cells. The deficiency of gzmA and gzmB in individual mice was confirmed by intracellular staining (Fig. 2⇓C), which also revealed that even substantially increased gzmA expression by perf-deficient CD8 T cells did not rescue the deficiency of these T cells to eliminate targets in vivo. Furthermore, virus-immune mice deficient in the gzm-activating enzyme cathepsin C (dipeptidyl-peptidase I), who express gzmA and B zymogen proteins but have been shown in vitro to express only residual gzmB enzymatic activity (26), were also as efficient in eliminating target cells as WT mice (Fig. 2⇓D).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

In vivo killing by antiviral Tc cells is dependent on perforin, but not gzmA or gzmB. WT or gene KO mice were infected with ECTV and 6 days later injected with cognate (EV-K^b, CFSE^high) and control (control peptide, CFSE^low) splenocytes from naive Fas^−/− mice. At the indicated time points spleens were removed and adoptively transferred target cells were determined by FACS. A, Relative elimination of cognate targets 150 min posttransfer. B, Proportion of splenocytes binding tetramers comprising the ECTV Tc cell determinant (EV-K^b) used to label cognate target cells. C, GzmA (open bars) and gzmB (filled bars) protein expression by splenic CD8⁺ cells. D, Relative elimination of cognate target cells 60 min posttransfer in WT or gene KO mice, including mice deficient in the gzm-converting enzyme cathepsin C (B6.CatC). Data shown are means with SD from three mice per group and are representative of at least three experiments.

It is important to note that only a small fraction (typically 6–10% at 1 h posttransfer) of adoptively transferred splenocytes are found in the recipient spleen at any one time (Ref. 27 and our unpublished data), and we do not know whether the bulk of donor splenocytes had become entrapped in other tissues during circulation, extravasated into peripheral tissues, or died. It is also unclear whether any of the following are true: 1) these elimination processes occur between spleen-resident Tc cells interacting with target cells that have lodged in the spleen; 2) spleen-resident Tc cells screen circulating cells and eliminate targets as they pass through the spleen; or 3) elimination occurs in a different organ altogether. Mathematical models have sought to address these questions (24, 25). However, these variables are likely to be the same for both WT and gzmA×B^−/− mice, because their overall organogenesis and lymphoid tissues are similar (6). We consistently found similar specific disappearances of viral peptide-pulsed targets in popliteal lymph nodes, suggesting that target cell elimination is either systemic or at least occurs evenly throughout the secondary lymphoid organs. Sometimes lungs and livers were also harvested and, where both target cell populations were detectable in sufficient numbers, the target cell ratios were similar to those in lymphoid organs (not shown).

No compensatory up-regulation of orphan gzm

Because other gzm might compensate for the loss of gzmA and gzmB in the gzmA×B^−/− mice, we determined the expression of several gzm by semiquantitative PCR, which revealed a similar expression profile between WT and gzmA×B^−/− mice (Fig. 3⇓). Only gzmC appeared to be expressed more strongly in WT mice, consistent with the reported “knock-down” effect of the gzmB deletion in the KO mice used (28). Thus, if anything, gzmA×B^−/− Tc cells appear to be more deficient in components of the granule exocytosis pathway than WT Tc cells, despite their equal capacity to eliminate cognate targets. However, it should be noted that, as shown for NK cells, mRNA expression may be an unreliable indicator of gzm protein expression (12).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

Expression of gzm and perf genes. CD8⁺ splenocytes from experiments in Fig. 2 were enriched by MACS. Total mRNA from 5 × 10⁵ cells was isolated and reverse transcribed, and the resulting cDNA was used in real-time PCR using TaqMan gene expression assays. Data shown are orphan gzm gene copy numbers relative (rel.) to copy numbers of GAPDH mRNA from two mice per strain.

Influenza virus-immune Tc cells also eliminate cognate target cells independently of gzmA and B

Finally, gzmA×B-independent in vivo killing was not specific to the ECTV- infection model studied, because influenza A virus-immunized gzmA×B^−/− mice displayed a similar cytotoxic competency when compared with influenza A-infected WT mice (Fig. 4⇓), although killing was less efficient overall than that found in ECTV-immune mice, consistent with the weaker antiviral Tc cell response induced in the former infection model. As seen in the ECTV infection model, the in vivo killing observed was not mediated by FasL but was dependent on perf. This suggests that antiviral gzmA×B^−/− Tc cells are able to eliminate specific targets in vivo as efficiently and rapidly as WT Tc cells and that Fas-mediated killing is not involved when perf is present, at least not within the time frame of observation.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

In vivo Tc cell activity in influenza-infected mice. Seven-day influenza virus-immune mice deficient in gzm, perf, or FasL (gld) were injected with target cells from naive Fas^−/− mice and pulsed with an immunodominant influenza-derived Tc cell determinant (NP-D^b) o/r a control peptide (EV-K^b). Four hours later recipient spleens were analyzed for specific loss of target cells. Data shown are from two (gld recipients) or three mice (all others) per group and representative of two experiments.

Concluding remarks

We provide evidence that neither gzmA nor gzmB is required for efficient and rapid perf-dependent elimination of cognate Tc cell targets in vivo. These observations present a conundrum in the context of previous in vitro observations, including our own, which until recently supported a model of gzm function where gzmA and B are required for optimal induction of perforin-dependent apoptosis. Although this gzmA/B dependency is variable to a limited degree, depending on the type of target cell investigated (Ref. 7 and manuscript in preparation), all assays have shown substantial kinetic defects in apoptosis without these two granzymes. Under the conditions described, lack of gzmA and B does not appear to be compensated in vivo by increased expression of any of the orphan gzm or by FasL-Fas interactions. It appears that Tc cell-delivered perf alone or in combination with other unknown factors may be sufficient to induce the efficient elimination of target cells in vivo. Thus, the mechanism(s) by which gzmA and gzmB contribute to recovery or protection from several virus infections and tumor models is still uncharacterized, but it appears not to be due to a lack of cell death induction per se. Other mechanisms, including direct antiviral effects, have been demonstrated or postulated for specific viral models (17, 29, 30, 31, 32).

Another possibility is that gzmA and B, while not necessary for cell death induction per se, may provide downstream signals to help establish or maintain an appropriate immune response via the type of cell death induced. An important immunological component that is not present in in vitro assays is the extremely efficient removal of dead or dying cells by the phagocytic system. Importantly, our observation that gzmA appears to be dispensable for cytotoxicity is in agreement with a recent report by Metkar et al. (17) that questioned the in vitro cytotoxicity of gzmA, attributing the cytotoxic potential shown by previous studies to contamination or the unphysiologically high gzmA concentrations used for in vitro assays. Instead, the authors found that gzmA delivered by Tc cells was able to induce IL-1β secretion from activated macrophages, suggesting a role in the inflammatory response to target cells attacked by Tc cells. We have preliminary data suggesting that the phagocytic response to targets killed in gzmA×B-sufficient and -deficient mice differs, and we are currently investigating the consequences of that difference.